Mechanisms of Development 69 (1997) 169–181

Identification of the vertebrate Iroquois family with overlapping expression during early development of the nervous system

Antje Bossea,*, Armin Zu¨lcha, May-Britt Beckera, Miguel Torresb, Jose´ Luis Go´mez-Skarmeta c,d, Juan Modolelld, Peter Grussa

aDepartment of Molecular Cell Biology, Max Planck Institute of Biophysical Chemistry, Am Fassberg, D-37077 Go¨ttingen, Germany bDepartamento de Immunologia y Oncologia, Centro National de Biotecnologia, Universidad Autonoma Cantoblanco, E-28049 Madrid, Spain cLaboratorio de Biologia del Desarollo Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile dCentro de Biologia Molecular Severo Oatoa, C.S.I.C. and U.A.M., E-28049 Madrid, Spain

Received 25 August 1997; revised version received 6 October 1997; accepted 7 October 1997

Abstract

In Drosophila the decision processes between the neural and epidermal fate for equipotent ectodermal cells depend on the activity of proneural . Members of the Drosophila Iroquois-Complex (Iro-C) positively regulate the activity of certain proneural AS-C genes during the formation of external sensory organs. We have identified and characterized three mouse Iroquois-related genes: Irx1, -2 and -3, which have a homeodomain very similar to that of the Drosophila Iro-C genes. The sequence similarity implies that these three genes represent a separate homeobox family. All three genes are expressed with distinct spatio/temporal patterns during early mouse embryogen- esis. These patterns implicate them in a number of embryonic developmental processes: the A/P and D/V patterning of specific regions of the central nervous system (CNS), and regionalization of the otic vesicle, branchial epithelium and limbs.  1997 Elsevier Science Ireland Ltd.

Keywords: Iroquois; Prepattern gene; Homeobox; Mouse; Xenopus; Human; Neurogenesis; Otic vesicle; Branchial cleft; Limb

1. Introduction selection process requires a lateral inhibitory signaling mediated by the neurogenic genes. The key neurogenic During development of the mammalian nervous system, genes such as Delta, Notch, Suppressor of Hairless neuronal precursor cells within the ventricular zone prolif- [Su(H)], and the Enhancer of split complex [E(Spl)-C] erate and differentiate into the different cell types of neu- encode that restrict the adoption of a neural fate rons and glia. At the molecular level, however, little is to one or a few cells within the proneural cluster, reviewed known about the genes which direct neural cell commit- by (Muskavitch, 1994; Parks et al., 1997). ment. Homologs of the Drosophila proneural genes have been In Drosophila, genes playing critical roles during neural identified in many vertebrate species, and for some of them cell lineage determination have been identified, composing proneural activities have been reported (Allende and Wein- a network of proteins encoded by proneural and neurogenic berg, 1994; Ferreiro et al., 1994; Guillemot, 1995); for genes, for reviews see (Campuzano and Modolell, 1992; reviews see (Kageyama et al., 1995; Lee, 1997). An exam- Ghysen et al., 1993). Proneural genes, such as the four ple of this is Mash1, which represents a mouse homolog of members of the achaete-scute complex (AS-C), mediate the AS-C genes, with an essential role during the formation the delineation of groups of ectodermal equivalent cells of autonomic and olfactory neurons (Guillemot and Joyner, (proneural cluster) and confer them the capacity to become 1993; Sommer et al., 1995). In vertebrates, increasing evi- neural precursors, reviewed by (Jan and Jan, 1994). Within a dence emphasizes the function of the neurogenic and pro- proneural cluster not all cells generate neurons, and this neural regulatory network to select individual cells from equivalent groups for specific fates (Bellefroid et al., * Corresponding author. 1997; Go´mez-Skarmeta et al., 1997). The identification of

0925-4773/97/$17.00  1997 Elsevier Science Ireland Ltd. All rights reserved PII S0925-4773(97)00165-2 170 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 genes like Hes and Id representing the murine counterparts and low stringent hybridization conditions. The cDNA of Hairy and emc, two negative regulators of A/S-C genes, sequence analysis confirmed that three members of the as well as the isolation of mammalian neurogenic genes mouse Iroquois homeobox family were cloned, which we supports this (Moscoso del Prado and Garcia-Bellido, named Iroquois-homeobox-l, -2 and -3 (Irx1, -2 and -3) (Fig. 1984; Chitnis and Kintner, 1996). 1A). In order to examine whether the evolutionary conserva- Alignments of the deduced homeodomains and flanking tion of the proneural genes also includes some of their posi- sequences of Irx1, -2 and -3 show regions of high amino tive upstream regulators we focused on the Drosophila acid sequence identity (82–92%) restricted mainly to the prepattern genes of the Iroquois complex (Iro-C). Two of homodomains (Fig. 1B) and to some amino acids (up to these homeobox genes araucan (ara) and caupolican 20) in the carboxy-flanking region (not shown). (caup), were recently reported to positively control the Close inspection of the mouse Irx amino acid sequences site specific activity of ac and sc (two genes of the AS-C) revealed conservation of several features in addition to their in sensory organ proneural clusters (Go´mez-Skarmeta and homeodomains: the occurrence of potential phosphorylation Modolell, 1996; Leyns et al., 1996). Several lines of evi- sites for mitogen-activated kinase (MAPK) and dence indicate that the members of the Iro-C are required some acidic regions downstream of the homeodomain. for the initial activation of ac and sc in certain sensory organ Such acidic regions may be involved in DNA-binding proneural clusters. In loss-of-function Iro-C mutants the which has already been shown for flanking regions of lateral sensory organs on the notum of the flies fail to other homeodomain proteins. Furthermore, a nine amino form, accompanied by a loss of ac and sc gene expression acid motif in the C-terminal portion without a known func- (Go´mez-Skarmeta and Modolell, 1996; Leyns et al., 1996) tion was found to be conserved between Irx3 and Irx1 and and reviewed by (Vervoort et al., 1997). Furthermore, there ara, caup and mirror (not shown). This motif has not been is molecular evidence to show that Iro proteins are able to found yet in Irx2, but it may be present in uncharacterized bind A/S-C enhancer sequences and that this binding is parts of the cDNA. necessary for ac and sc expression (Go´mez-Skarmeta and The highest degree of amino acid similarity among the Modolell, 1996). Recent data imply the identification of a different mouse homeodomain sequences was observed further member of the Drosophila Iroquois family: mirror between Irx3 and Irx1 (Fig. 1A,B). (mrr), another homeodomain protein which is very similar The amino acid sequences of the three murine Irx home- to ara and caup, exhibits defined activity during eye devel- odomains have 92–95% identity with Drosophila ara, caup opment (McNeill et al., 1997). and mrr (Fig. 1B). This strongly supports the idea that the The evolutionary conservation of at least some members isolated mouse homeobox genes represent the mouse homo- of the genetic regulatory hierarchy that establishes neural logs of the Drosophila Iro-C genes. A Lysine at position 22 fate during embryonic development encouraged us to screen of the fly homeodomains represents a unique difference for vertebrate Iroquois homologs. compared with vertebrate Iroquois homeodomains (Fig. In this study we present three members of the mouse 1A). The homeodomain of one of the three so far isolated Iroquois (Irx) gene family, referred to as Irx1, -2 and -3. Xenopus homologs of Iroquois-like genes, Xiro3 (Bellefroid They share 92–95% amino acid identity in the homeodo- et al., 1997) has 100% similarity to the homeodomain of main with the Drosophila Iro-C genes. All three genes are Irx3. Additionally, the 5’ regions of Irx3 and Xiro3 have not only expressed in overlapping patterns, but also in spe- high sequence similarity. However the homeodomain of cific patterns within the developing nervous system. The Irx1 and Irx2 show only 84 and 87% identity, respectively, sequence conservation between Irx1, -2 and -3 to the Dro- to that of Xiro3. The homeodomains of the Xenopus Iro- sophila Iro proteins and their gene expression patterns dur- quois-like proteins Xiro1 and Xiro2 differ only in one amino ing embryogenesis indicate that Irx1, -2 and -3 might be acid from the murine homeodomains of Irx1 and Irx2, involved in early determination processes during nervous respectively (Go´mez-Skarmeta et al., 1997). system development. This finding suggests the possibility The mouse Irx homeodomains have significant similarity that the neurogenic pathway appears to be conserved in (90–100%) to seven human EST (expressed sequence species which are very distantly related, such as the fruitfly tagged) sequences available in the ATCC-database (Fig. and the mouse. 1A,B). These human Iroquois-like sequences also show a significant similarity (90–98%) to each other. One of the seven human Iroquois-like sequences (IRX1) has the same 2. Results homeodomain amino acid sequence as Irx1 and Irx3, and the Xenopus Iroquois-like gene Xiro3. However, flanking 2.1. Characterization of the mouse Iroquois gene family regions of the homeodomain of this human IRX1 sequence related to the Drosophila Iro-C differs more to that of Irx1 and Irx2 than to Irx3. An alanine residue at position nine of the recognition We isolated three novel mouse genes from a mouse helix (position 50 of the homeodomain) is characteristic cDNA library using a Drosophila ara homeobox probe for all members of the Iroquois protein family. No other A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 171

Fig. 1. (A) Comparison of the putative Irx1, -2 and -3 homeodomain sequences with other Iroquois-related homeodomain sequences. Bold represents amino acid identity throughout the whole Iroquois family. Capital letters in parentheses denote the following: D, Drosophila (Go´mez-Skarmeta and Modolell, 1996; McNeill et al., 1997); M, mouse (EMBL Nucleotide Sequence Database accession numbers for Irx1, -2 and -3 genes are: Y15002, Y15001 and Y15000, respectively); H. human. The amino acids are numbered, starting at 1 from the first amino acid of the homeodomain. Roman numbers on top represent the helices of the homeodomain. The Iroquois family of homeobox genes encodes proteins characterized by the presence of an alanine residue at position 50 of the homeodomain. Irx1, Irx3, and five of seven human sequences lack four amino acids in front of the homeodomain, in comparison with Drosophila and the remaining two human sequences. Sequence inconsistencies within the homeodomain of the human IRX6 derived from the database are depicted as an X. (B) Pairwise comparisons of percentage amino acid identity in the conserved homeodomains among Drosophila, mouse and human Iroquois family members. We assumed the not determinable amino acids (X) in IRX6 as conserved in comparison with the other Iroquois homeodomains. known homeodomain has an alanine at this specific posi- lated from E10.5–E17.5 mouse embryos were used to deter- tion. This alanine residue is placed in the major groove of mine the sizes of Irx1, -2 and -3 transcripts. the homeodomain and it is a major determinant of DNA- The Irx1 probe identified a 2.5 kb transcript (Fig. 2A). binding specificity (Treisman et al., 1989). The Irx2 probe detected two distinct transcripts of 2.1 and Taken together, the structural similarities between the 2.9 kb, respectively, (Fig. 2B). The Irx3 probe detected two mouse, Drosophila, Xenopus and human Iroquois-like transcripts of 2.3 and 2.8 kb, respectively, (Fig. 2C). These genes suggest that they constitute members of a new family data confirmed that the different cDNAs correspond to three of homeobox containing genes. different genes.

2.2. Detection of Irx1, -2 and -3 transcripts by Northern 2.3. In situ analysis of Irx1, -2 and -3 expression during blot analysis early embryogenesis (E6.5–E7.5)

Northern blots with total and polyA + RNA samples iso- We compared the expression of the mouse Irx1, -2 and -3 172 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181

Fig. 2. Northern blot analysis of Irx1, -2 and -3 in post-implantation embryos. The size of the transcripts are indicated on the left and were estimated by comparison with the RNA ladder from Boehringer (not shown). The lane (A) represents a Northern blot of 10 mg total RNA isolated from E14 and hybridized with a 400 bp Irx1 specific probe excluding the homeobox. Only one band was visible after hybridization with Irx1 corresponding to a molecular weight of 2.5 kb. (B) The Irx2 probe (1.4 kb) encoding parts of the homeobox and adjacent sequences detected two transcripts of 2.1 and 2.9 kb. (C) Three micrograms of polyA + RNA from E14.5 was hybridized with a 1.1 kb Irx3 probe comprising the 3′ region of the homeobox and adjacent sequences. This Irx3 probe detects two bands of 2.3 and 2.8 kb. The size differences between the transcripts and the cDNA fragments indicate that the Irx1 and Irx2 cDNAs are incomplete at their 5′ ends. genes during embryonic development between E6.5 and the mesodermal wings overlying the presumptive head- E10.5 by whole mount in situ hybridization. fold region (Fig. 3D). In contrast, Irx3 signals are confined Irx3 is the earliest of the Iroquois-family genes to be to the ectodermal layer where they represent two expres- expressed. Transcripts of Irx3 are first found at E6.5 in the sion domains in the margins of the neural plate (Fig. 3F). extraembryonic portion of the egg cylinder, at high levels in Irx2 expression is not yet detectable at these stages (Fig. the ectoplacental cone and in the lining of the ectoplacental 3B). cavity (data not shown). All the structures expressing Irx3 are trophectoderm derivatives. The chorionic ectoderm 2.3.1. All three Irx genes display distinct expression which develops from the ectoplacental cone also shows patterns during neural tube closure Irx3 transcripts at mid-gastrulation (Fig. 3C). At E8.5, Irx3 mRNA is found dorsolaterally in the neural The embryonic expression of Irx3 starts at E7.5 as two folds of the prospective mesencephalon, within the rhom- symmetrical patches in the anterior region of the embryo bencephalon mostly in the precursor regions of rhombomere (Fig. 3C). At the same time the first Irx1 expression is 1 and -3, and additionally in the rostral part of the closing seen in comparable regions of the embryo (Fig. 3A). Trans- neural plate (Fig. 3I,L). During the ongoing development, verse sections revealed that Irx1 transcripts are restricted to Irx3 signals extend further towards the caudal region of the

Fig. 3. Expression of mouse Irx genes at mid-gastrulation stage E7 5 up to early neurulation at E8.5 (10–12 somites). Anterior is to the left. (A,B,C) lateral views on E7.5 embryos. (A) shows expression of Irx1 only in the embryonic portion. (C) shows Irx3 expression in extraembryonic and embryonic parts of the egg cylinder. At this stage, no transcripts of Irx2 are detectable (B). (D,F) transversal sections of the E7.5 embryos in (A,C). The plane of sections are illustrated (E). (F) Irx3 expression within the E7.5 embryo is confined to the ectodermal layer. (D) expression of Irx1 is restricted to the mesodermal wings. Expression of Irx genes in E8.5 embryos (G–L). (I,G) lateral and (H) dorsolateral views of E8.5 embryos. (G) shows expression of Irx1 in the presumptive midbrain and in the anterior foregut. Initial pre-rhombolnere 4 (pre-r 4) specific expression of Irx2 is seen in (H). The rhombomere identification was possible in respect to the preotic sulcus. (I) represents Irx3 expression in the presumptive midbrain, hindbrain and in the anterior spinal cord as well as in the anterior foregut. (L) transverse section at the hindbrain level of the embryo in (I) revealing the Irx3 expression in the neuroepithelium of the presumptive hindbrain as well as in the epithelium of the branchial pouches. (J) saggital section of the embryo in (G) which reveals the Irx1 transcripts in the presumptive hindbrain and in the posterior spinal cord where it is already closed. In (J), the contour of the embryo has been partially delineated to make the identification of the labeled regions easier. (K) transverse section at the hindbrain level of a slightly older (E8.75) embryo than the one in (H) with Irx2 expression limited to the neuroepithelium of the presumptive hindbrain at the level of pre-r 4. FB, forebrain; bpo, branchial pouches; ect, embryonic ectoderm; exe, extraembryonic ectoderrn; FG, foregut; H. heart; HB, hindbrain; MB, midbrain; mes, mesoderm; pos, preotic sulcus; rh4, rhombomere 4; SP, spinal cord.

Fig. 4. Overview of the Irx expression profiles at E9.5 (A,B,C) and E10.5 (D,E,F) embryos. The embryo in (A) was hybridized with Irx1 showing expression in the midbrain, hindbrain, epithelium of the first and second branchial arches and foregut. (B) depicts an embryo with Irx2 expression in the midbrain, hindbrain, spinal cord, epithelium of the first and second branchial arches and foregut. (C) shows Irx3 expression in the midbrain, hindbrain, entire spinal cord, epithelium of the first and second branchial arches, cephalic mesenchyme, lateral plate mesenchyme and foregut. At E10.5 mid-gestation stage, (F) the Irx3 expression extends within the posterior forbrain up to the hypothalamic region (arrowhead in F). Irx1 and Irx2 have their rostral limit of expression in dorsalthalamic regions (D,E). The cephalic mesenchyme expression domain of Irx3 at E10.5 includes the nasal pits (F). Weak expression of Irx1 and Irx2 within the developing limb starts at E10.5: (D) Irx1 in the region were the AER arises and (E) Irx2 in the dorsal part. FG, foregut; bare, epithelium of the branchial arch; cem, cephalic mesenchyme; HB, hindbrain; lpm, lateral plate mesoderm; MB, midbrain; HB, hindbrain; PT, pretectum; som, somite; SP, spinal cord. A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 173 174 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 neural tube. Within the neural tube, Irx3 transcripts are istic patterns of their early expression are maintained confined to the ventricular site (close to the lumen) of the (Fig. 4D,E,F). The rostral boundary of Irx1 and Irx2 expres- alar plate, excluding the roof plate and the most dorsal sion extends to the presumptive domain of the zona limitans region (Fig. 7C). intrathalamica (Fig. 4D,E). Only Irx3 shows expression Irx1 and Irx2 transcript distributions within the develop- in the basal plate of the diencephalon (arrowhead in ing nervous system appear more restricted in comparison to Fig. 4F). Furthermore, Irx3 expression in the mesencepha- Irx3. The most prominent expression of Irx1 in the brain is lic roof exhibits a posterior-to-anterior gradient (Fig. confined to the dorsolateral walls of the mesencephalon 4F). (Fig. 3G). Sagittal sections also revealed a weak expression Additionally, in the spinal cord Irx3 transcripts are loca- in the rhombencephalon and within the posterior region of lized in cells of the ventricular layer and in the motoneuron the neural tube (Fig. 3J). At this stage, Irx2 expression column of the basal plate (Fig. 7E). The alar plate reveals appears for the first time in the rhombencephalon, specifi- also two restricted areas in the marginal layer. cally active in the presumptive region of future rhombomere In these stages, Irx1 expression extends into the entire 4 (Fig. 3H). ventral spinal cord, but the strongest expression remains in the most caudal region (arrowhead in F. 4D). From this 2.3.2. During neurogenesis (E9.5–E10.5) Irx1, -2 and -3 time onwards Irx1 expression domain is restricted to the are predominantly expressed along the anteroposterior ventral spinal cord. axis of the CNS At E9.5 all three Irx genes display restricted patterns of 2.3.3. Complementary expression of Irx1, -2 and -3 during expression along the A/P axis of the developing neural tube early development of the inner ear with a common rostral limit at the level of the pretectum Starting at the otic vesicle stage, there is a remarkable (Fig. 4A,B,C). All Irx genes are strongly expressed in the regionalized expression pattern of all three Irx genes in the tectum of the mesencephalon, whereas in the tegmentum developing inner ear (Fig. 6). Transverse sections of E9.5 only Irx3 and Irx1 transcripts are found (Fig. 4A,B,C). embryos at the level of the hindbrain reveal the expression The hindbrain also exhibits A/P-specific expression of the of Irx1 within the lateral wall of the otic vesicle (Fig. 6A). three Irx genes: Irx3 is expressed throughout the entire hind- This region partially overlaps with the Irx2 expression brain while Irx1 and Irx2 are restricted to specific rhombo- domain in the lateroventral part of the otic vesicle (Fig. meres (Fig. 4A,B,C). Sagittal sections reveal only a faint 6A,B). In contrast to Irx1 and Irx2, the distribution of Irx3 appearance of Irx1 mRNA within the dorsal portions of the transcripts is restricted to the ventromedial portion of the rhombomeres 2, 3 and 4 (Fig. 7B). At this stage Irx2 expres- otic vesicle adjacent to the hindbrain (Fig. 6A,B,C). The sion is found in rhombomeres 1–4 and 7–8 (Fig. 4B). It is dorsal region of the otic vesicle is devoid of any detectable worth noting that the border region between mesencephalon Irx transcripts. From E10.5 onwards, all three Irx genes are and rhombencephalon and the anterior prosencephalon do expressed in the entire otic vesicle. not show any Irx gene activity. The condensing mesenchyme on the ventral side of the In the rostral spinal cord, Irx3 transcripts are found in the otic vesicle also displays expression of Irx1, -2 and -3 alar and basal plate, but not in the roof- and floorplate (Fig. (arrowhead in Fig. 6). The surrounding mesenchyme is 5I,L). In caudal regions of the spinal cord Irx3 is expressed necessary for proper spatial development of the vestibular only in the alar plate (Fig. 5L). In contrast, Irx2 expression and auditory receptors, and for the formation of the carti- in the spinal cord is much weaker and confined to the alar lages which later surrounds the inner ear (Ruben et al., plate along the entire spinal cord (Fig. 5H,K). The expres- 1986). sion of Irx1 is only observed in the spinal cord posterior to the region of the developing hindlimb (Fig. 5J). 2.3.4. Specific expression of Irx1, -2 and -3 during limb At E10.5, the expression domains of all Irx genes extend development rostrally into the dorsal diencephalon, while the character- Only Irx3 is expressed in the prospective limb territories

Fig. 5. Vibratome transverse sections of embryos at E9.5 following Irx whole mount in situ hybridization. (A–F) Irx1, -2 and -3 expression in the hindbrain. Irx gene expression in the cephalic mesenchyme around the eye (F) and in cells which may participate in the trigeminal ganglia formation (A,F). Irx1, -2 and -3 are transcriptionally active in the foregut (G–I), in particular in regions which develop the lungbud (tracheal diverticulum) (J–L). Irx1 expression within the spinal cord is restricted to a caudal region (G,J). Weak Irx2 expression within the spinal cord is restricted to the alar plate (H). (I,L) shows Irx3 expression in the entire spinal cord. Irx1 (D,G) and Irx2 (B,E,H,K) are transcriptionally active in the surface ectoderm. FB, forebrain; bpo, branchial pouches; cem, cephalic mesenchyme; ect, surface ectoderm; FG, foregut; HB, hindbrain; lpm, lateral plate mesoderm; MG, midgut; SP, spinal cord; tra, tracheal diverticulum.

Fig. 6. Regionalized Irx expression in the otic vesicle. Whole mount in situ hybridization of transverse sections (A,B,C) at the hindbrain level of the E9.5 embryos in (D,E) which are depicted in dorsal views. Irx1 (A) and Irx2 (B) overlap in lateroventral regions; Irx1 expression extends further laterally than Irx2. (C) Irx3 expression is restricted to the otic vesicle neuroepithelium adjacent to the hindbrain. HB, hindbrain; ov, otic vesicle. A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 175 176 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 of the lateral plate mesoderm from E9.0 onwards (Fig. 4C). (Fig. 7D). This expression extends distally as the limb From E10.5 onwards, the Irx genes exert distinct patterns grows and maintains the gradient. From E10.5 onwards of expression in the developing limb buds (Fig. 4D,E,F). Irx1 and Irx2 start to be expressed in the limb bud Transcripts of Irx3 exhibit a gradient of expression along (Fig. 4D,E). Irx1 appears to be very faintly expressed in a the dorsoventral and the proximodistal axes with their region where the apical ectodermal ridge (AER) arises highest level in the proximodorsal margin of the limb (Fig. 4D). Only a weak Irx2 expression is detectable in

Fig. 7. Special aspects of Irx expression. (A) Transverse section of an E9.0 embryo showing Irx3 expression in the notochord and neural tube. (B) Midsagittal section of an E9.5 embryo with Irx1 expression restricted to dorsal portions of the rhombomeres 2, -3 and -4, the midbrain and the ectodermal layer of the first branchial arch. (C) Transverse section of an E8.5 embryo shows Irx3 positive ventricular cells close to the lumen of the neural tube. (D) Section of a limb at E10.5 with Irx3 transcript accumulation in proximodorsal regions. (E) Transverse section of an E10.5 spinal cord showing Irx3 labeling in the ventricular layer and in the motoneuron columns of the basal plate. Arrowhead marks the Irx3 expressing cells in the marginal layer of the alar plate. Scale bars represent ~50 mm in (A,C) and ~250 mm in (B,D,E). APL, alar plate; BAP, basal plate; bare, epithelium of the branchial arch; d, dorsal; dt, distal; H. heart; HB, hindbrain; Hl, hindlimb; MB, midbrain; mot, motoneuron; nch, notochord; NE, neuroepithelium; ov, otic vesicle; p, proximal; som, somites; v, ventral.

Fig. 8. Schematic summary of the comparative expression analysis of Irx1, -2 and -3 between E9.5 and E10.5. Hatched areas delineate mesodermal derivatives. Endodermal derivatives are depicted with black dots. Irx expression within the CNS is shown colored. The three genes are expressed along the CNS with a rostral limit at the dorsal thalamus except for Irx3 which extends to the hypothalamus. I–IV, branchial arches I–IV; DT, dorsalthalamus; ET, epithalamus; Fl, forelimb;; H. heart; HB, hindbrain; Hl, hindlimb; Hyt, hypothalamus; L, lungbud; MB, midbrain; PT, pretectum; som, somites; VT, ventralthalamus. A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 177 the dorsal portion of the limb bud (Fig. 4E). At E11.5 tran- 3.1. Conservation of the predicted protein sequences of the scripts of Irx1 appear strongly in the forelimb digits (not murine Iroquois-like genes is mainly restricted to their shown). homeodomains

2.3.5. Irx1, -2 and-3 expression in other tissues during The primary protein sequences of Irx1, -2 and -3 reveal a early stages of organogenesis high degree of identity (92–100%) within their homeodo- Both Irx1 and Irx3 are expressed in two symmetrical mains. This extreme amino acid sequence conservation in patches in the embryonic mesodermal layer, with Irx3 Iroquois-like proteins was also found when compared to expression being very faint. They begin to express at stages such evolutionary distant species as Drosophila, Xenopus E7.5 and E7.75, respectively (Fig. 3D). However, only Irx3 and human (up to 100% amino acid identity between mouse is detected in the notochord at stage E9.0 (Fig. 7A). At E9.5 and human). An alanine residue at position 9 of the recogni- transcripts of Irx1 and Irx3 and later also of Irx2 (at E10.5) tion helix is a characteristic feature for the Iroquois family are found in the cephalic mesenchyme around the optic of homeodomains. Due to the high sequence similarity and vesicle which is involved in the formation of the outer the characteristic expression patterns we consider the mouse layers of the eyes, the eye muscles and the lachrymal glands Iroquois-like genes as a separate homeobox family as it was (Fig. 4A,C,E). Around E10.5, the Irx3 signals in the proposed by Go´mez-Skarmeta and Modolell (1996). Preli- head mesoderm extend into the nasal pits (Fig. 4F). Further- minary data indicate that the murine Iroquois-like gene more, all three genes are found in putative cephalic family is larger than the three members presented (unpub- mesenchymal cells which in time may participate in trigem- lished observations). inal and fascio-acustic ganglia formation (Fig. 5A,F). At later stages the Irx genes are transcriptionally active in the 3.2. Involvement of Irx genes during neurogenesis V and VII–VIII cranial ganglia (not shown). The cranial ganglia are known to have two embryonic origins: the The finding that Drosophila Iro-C members regulate cephalic neural crest and the epibranchial placodes. From proneural genes suggests that the murine Irx genes could E9.5 onwards, the branchial clefts including the regions of play a similar role during mouse neurogenesis. In Droso- the branchial placodes, reveal different restricted expression phila, the genes of the Iroquois-complex are expressed in patterns of all three genes in distinct ectodermal regions the head of the embryo, epidermis and in broad territories between the mandibular, maxillar and hyoid swellings within the imaginal wing disc that include the proneural (Fig. 4 A–F), schematically summarized in Fig. 8. How- clusters which give rise to the sensory organ precursor ever, no Irx expression is detectable in migrating neural cells (Go´mez-Skarmeta and Modolell, 1996). However, crest cells. Later on, the external ears (derivatives of the the exact pattern of expression in the head of the fly embryo epibranchial placodes) show expression of Irx1, -2 and -3 has not yet been described. (not shown). Three Xenopus homologs of the Iroquois family are Beginning with E9.5, Irx1 and Irx2 exhibit expression in found to be expressed in the neuroectoderm and anterior the superficial ectoderm surrounding the body, with Irx1 mesoderm, overlapping Xash3 expression (Bellefroid et transcripts being restricted to the ectoderm at the level of al., 1997; Go´mez-Skarmeta et al., 1997). Their recent data the hindbrain (Fig. 5D,E,G,K). imply that the Iroquois-like genes in Xenopus promote neu- From E8.5 onwards, Irx1 and later all three Irx genes rogenesis upstream of Xash3 and other proneural genes. (E9.5) are expressed in the region of the foregut, which The early and transient expression of the mouse Irx genes will form the pharynx and the lung bud (Fig. 3G). This during initial stages of neural development represents a expression domain is maintained at later stages in the common feature of the vertebrate and invertebrate Iroquois epithelial layer of the bronchiae (Fig. 5). family members. Irx3 is expressed in the neuroectoderm at a At E10.5 only Irx1 expression appears in the somites and time when the neural plate begins to form (E7.5). Whether becomes later restricted to the myotome (Fig. 4E and sec- this similarity in the chronology of gene activity indicates tion not shown). that this part of the network of interacting genes is also conserved through evolution remains to be investigated. Furthermore, the localization of Irx3 transcripts in the 3. Discussion ventricular zone of the alar plate of the spinal cord contain- ing uncommitted neural precursor cells suggests that Irx3 We have characterized three mouse homeobox genes could play a role in the development of sensory neurons. Irx1, -2 and -3 related to genes of the Drosophila Iro-C Neurogenesis in mammals requires progressive regiona- locus. These mouse genes code for proteins exhibiting a lization along the A/P and D/V axes resulting in regions very high homeodomain sequence conservation as their with distinct molecular traits. Many transcription factors Drosophila counterparts. Their expression precedes and and regulatory molecules participate in these patterning pro- overlaps the expression of distinct proneural genes in the cesses as shown for the Hox, Pax, Nkx and POU gene mouse. families (Guillemot, 1995). For example, evidence has 178 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 been presented for the involvement of Hox genes in the reviews see (Mansouri et al., 1996; Tanabe and Jessell, specification of hindbrain identities, for reviews see 1996). (McGinnis and Krumlauf, 1992; Keynes and Krumlauf, In conclusion, the Irx genes may participate in the signal- 1994). During vertebrate hindbrain development, 8 meta- ing cascade during the regionalization along the neuraxis. meric units, termed rhombomeres, appear. Only a few Their distinct spatio/temporal expression profile between genes are described which display a rhombomere specific E9.5 and E10.5 is schematically summarized in Fig. 8. activity such as Hoxbl, the most 3’ member of the HoxB The overlap in expression patterns of the three Irx genes group and the zinc-finger Krox20 (God- during early embryogenesis raises the possibility of func- dard et al., 1996); reviewed by (Lumsden and Krumlauf, tional redundancy of these genes. A systematic mutational 1996). analysis of single and compound mutants of the Irx genes, Interestingly, even prior to the formation of morphologi- together with the comparative expression studies as pre- cally recognizable rhombomeric units, Irx2 expression is sented here, will identify both the functions that are shared restricted to the hindbrain neural ectoderm giving rise to among the different members and the unique roles of each pre-rhombomere 4 (pre-r4). Irx3 signals accumulate in of them. pre-r1 and pre-r3. These specific expression patterns suggest that the Iroquois-like genes may be involved in rhombo- 3.3. Complementary expression of the murine Irx genes in mere territory delineation and could be useful markers for the otic vesicle suggests a role in pattern formation during rhombomere identities. inner ear development At E9.5 all three Irx genes exhibit an anterior expression border at the pretectum. One day later, this border expands The otic vesicle is formed by a process of condensation of to the dorsal thalamus. Only Irx3 extends its expression also ectodermal cells into the otic placode, which then invagi- into the basal plate of the diencephalon. nates into the mesenchyme lateral to rhombomere 5 and -6. Within the mesencephalic roof Irx3 exhibits a rostrocau- The otic vesicle gives rise to the inner ear which is also dal gradient of expression. The earliest genes reported so far described as the membranous labyrinth and contains the with expression in a gradient-like manner are the sensory organs for hearing and balance. Between E9 and genes, reviewed by (Lumsden and Krumlauf, 1996). Knock- E10, the otic vesicle becomes regionally determined, out experiments demonstrated that Enl is indeed involved in although its neuroepithelial walls still look homogenous establishment of midbrain polarity, reviewed by (Joyner, (Ruben and Rapin, 1980; Swanson et al., 1990). 1996). The expression patterns of otic vesicle regionalizing The Irx genes may also participate in the A/P regionali- genes may indicate the future cell fate of the developing zation of the spinal cord. While Irx3 and Irx2 were active inner ear, for review see (Fekete, 1996; Rivolta, 1997). along the entire A/P axis of the spinal cord at E9.5, tran- Inactivation of a few of these genes lead to inner ear mal- scripts of Irx1 were limited to the region posterior to the formations which help to determine the region of the otic hindlimb bud. One day later Irx1 expression extends ros- vesicle which gives rise to the individual structures of the trally, but still exhibits its most intensive expression in the inner ear (Mansour et al., 1988; Torres et al., 1996; Hadrys tail bud region. et al., 1997). Firstly, the dorsomedial aspect of the otic Moreover, Irx1, -2 and -3 display distinct patterns also vesicle extends to form the endolymphatic sac while its along the D/V axis of the developing CNS suggesting a ventral end differentiates into the cochlea. The utricular possible involvement in early neuronal regionalization. region descending from lateral and dorsal parts differenti- Irx3 is the only Irx gene with expression in the basal plate ates to form the three semicircular canals. These differentia- within the diencephalon, while both Irx1 and Irx2, are tion processes during inner ear formation, as in any other limited to the dorsal and ventral thalamus, which are alar organ system, most likely result from the temporal and spa- plate derivatives. Within the mesencephalon as well as the tial expression of a selected array of genes. pretectum and tectum (the last two being alar plate descen- The complementary expression of the three Irx genes dants) all three Irx genes are expressed. The basal mesence- suggests that they together with other regulatory genes con- phalic domain, the tegmentum, expresses Irx3 and Irx1, but tribute to the development of inner ear components. not Irx2. The spinal cord also displays a distinct D/V pattern Specifically, the region of Irx1 activity in the lateral por- of Irx genes: Irx1 appears to be specific for ventral (basal) tion of the otic vesicle overlaps with the expression of the territory while Irx3 and Irx2 are restricted to the most dorsal homeobox gene Nkx5.1. Loss-of-function of Nkx5.1 results region. It is remarkable that the main Irx3 and Irx2 expres- in malformation of all three semicircular canals (Hadrys et sion domains in the neural tube occur in alar plate deriva- al., 1997). Thus, one can speculate that Irx1 also participates tives. in semicircular canal formation Irx2, with its transcriptional Candidates for determination of neuronal identities in activity in ventrolateral regions of the auditory vesicle, may the CNS are Pax3, Pax7 and Mash1 which are expressed be involved in cochlea development. Irx3 signals were con- during dorsal cell differentiation, and also Pax6 and Neuro- fined to the otic vesicle neuroepithelium adjacent to the genin which are expressed in medial/ventral regions, for hindbrain, similar to the expression of Pax2. Since mice A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 179 in loss-of-function studies of Pax2 exhibited cochlea and arises, while later Irx1 expression was located in the digits ganglia defects (Torres et al., 1996) we suppose that Irx3 (not shown). The AER is involved in the separation of the also participates in cochlea development and formation of digits (Williams et al., 1989). Experiments removing the the VIII ganglion. Moreover, Irx3 and Pax2 may be func- AER led to truncated digits. Thus, Irx1 may contribute to tionally related because their expressions overlap comple- the process of digit formation. tely at the onset of Pax2 activity in late primitive streak stages (Rowitch and McMahon, 1995). 3.5. Irx genes exhibit expression in concordance with It is known that inductive processes mediated by rhom- Mash1 in the CNS and elsewhere bomere 5 and -6, which are required for the formation and subsequent development of the auditory and balance system In Drosophila it has been found that the Iro genes posi- in vertebrates (Wilkinson et al., 1988; Noden and Van de tively control some of the A/S-C genes (Go´mez-Skarmeta Water, 1992). Fgf3 (Int2) whose expression occurs in a and Modolell, 1996). region proposed as an inductive area for the otic vesicle is Mash1 represents the only mammalian homolog of the presumed to encode a diffusible factor produced by the AS-C genes with transcriptional activity during embryonic hindbrain (Wilkinson et al., 1989). In loss-of-function nervous system development. In analogy to the functional Fgf3 mutants the otic vesicle and tail structures fail to cascade in the fly, Mash1 could be a potential target gene of develop properly (Mansour et al, 1988). However, in these mammalian Irx genes. Mash1 is expressed in proliferating mice the induction of the otic vesicle itself is not affected, precursors in CNS and autonomic NS during prenatal mouse rather, its subsequent development. development (Guillemot and Joyner, 1993; Lo et al., 1994). According to these observations it is of particular interest At E9.5 the distribution of Irx1, -2 and -3 transcripts com- that all three Irx genes are strongly expressed in rhombo- pletely match the Mash1 expression within the CNS in the mere 5 and -6 when the initiation of the otic placodes takes mesencephalic roof, hindbrain and spinal cord. One day place, indicating a possible involvement of Iroquois-like later, the Irx genes and Mash1 are coexpressed in the dien- genes in morphogenetic induction events during the forma- cephalon, tegmentum and in certain sensory ganglia. In tion of the otic vesicle (at about E8.0–E8.5). addition, Mash1 expression can be correlated with Irx During embryonic inner ear development our expression gene activity in the developing lung. The spatial expression data imply two different involvements of the mouse Irx of Irx genes, which at least partially overlaps with Mash1 genes: firstly, during otic vesicle induction by the hindbrain, expression in the mouse embryo, suggests a possible con- and secondly, during otic vesicle patterning and differentia- servation of the regulatory genetic interaction from fly to tion. mouse. In particular, Mash1 starts to be expressed at E9.5, and 3.4. Specific Irx activity suggests roles in determination of thus differs from the onset of Irx expression (Guillemot limb territories and Joyner, 1993). As a possible downstream target, also the mouse homolog to Xash3 would be an interesting can- The limb bud emerges from thickened lateral plate meso- didate for a role in neuronal induction (Zimmerman et al., derm which condenses with its ectodermal covering. Mem- 1993). bers of the Fgf family are known to initiate limb bud Further upstream regulatory genes for the Drosophila outgrowth (Dealy et al., 1996; Ohuchi et al., 1997). Irx3 Iro-C locus have been reported (Go´mez-Skarmeta and activity at E9.0 in the lateral plate mesoderm (prior the Modolell, 1996). Thus, a zinc-finger domain containing limb bud appearance) suggests a role during these early gene, cubitus interuptus (ci), functioning as a segment limb formation processes. polarity gene in Drosophila may regulate together with The limb as an asymmetric structure is defined on three dpp the activity of the Iro-C genes. Homologs have been axes: anterioposterior (A/P), proximodistal (P/D) and dor- identified in C. elegans (tral), mouse (Gli1-3) and human soventral (D/V). A series of factors participating in axial (Vortkamp et al., 1991; Schimmang et al., 1992; Hui et al., determination of the limb bud have been proposed, such 1994). as Wnt7a which is involved in D/V patterning, Shh which Interestingly, our mouse Irx expression data reveal sev- is specifying posterior identities, members of the Fgf family eral parallels to the expression of putative upstream genes, for distal outgrowth, LMX1 for dorsalization and En1 which the members of the Gli-family. The initial expression of the is essential for limb ventralization (Dealy et al., 1996; Ohu- Gli genes is detected during gastrulation in the overlapping chi et al., 1997). Another important group of genes are the ectoderm and mesoderm and later with distinct patterns in HOX genes homolog to the Drosophila HOM-C (Krumlauf, the developing neural tube similar to the Irx expression. 1994). Sections of embryos hybridized with Irx3 suggest its Moreover, the Gli genes show also a D/V patterning in the possible involvement during establishment or maintenance developing neural tube like the Irx genes. A naturally occur- of the D/V and P/D axes. ring Gli3 mutation, the mouse extra-toes mutant, shows The initial Irx1 expression in the limb bud from E10.5 neural tube closure defects and skeletal malformations onwards was limited to the region from which the AER (Hui et al., 1994). Mutations in the human Gli3 gene lead 180 A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 to the developmental disorder GCPC (Greig-Cephalo-Poly- 4.3. In situ hybridization syndactylie-syndrome) which has similar limb and cranio- facial malformations (Vortkamp et al., 1991). The cloning of the murine Irx members from a E8.5 In conclusion, the sequence similarity of mouse Irx genes cDNA library implies expression during early stages of combined with their pattern of expression during develop- embryonic development. Whole mount in situ hybridiza- ment suggests both a structural and a functional correlation tions (Wilkinson and Nieto, 1993) using in vitro probes with Drosophila Iroquois genes. This notion is further sup- deriving from regions outside of the conserved Irx homeo- ported by the similarity between the expression patterns of box were performed on mouse embryos from E6.5–E10.5. Irx genes and those of the putative mammalian homologs in For each gene we used at least two different probes between the Drosophila signaling cascade. 0.5–1.0 kb, and the detected patterns were identical. Sense RNA probes were used as negative controls in all experi- ments. Structures in the developing nervous system were 4. Experimental procedures named according to the atlas of (Alvarez-Bolado and Swan- son, 1996). 4.1. cDNA analysis

A homeobox derived 700-bp probe from the Drosophila Acknowledgements ara gene was used to screen a mouse E8.5 C57BL murine lgtlO poly (T)-primed cDNA library (kindly provided by We are especially grateful to Kamal Chowdhury for Dr. Brigid Hogan) according to (Oliver et al., 1995). A total advice on various experimental procedures; Anastasia Stoy- of 106 plaques were screened and 13 independent phages kova, Barbara Meyer, Lydia Lemaire, Kenneth Ewan and were isolated. Sequence analysis revealed that the 13 cDNA Eva Bober for stimulating discussions on the manuscript. fragments contained three diverged murine Iroquois related We wish to thank also Ralf Altscha¨ffel for excellent photo- genes. graphic work and ClausPeter Adam for the illustration in Four cDNA clones represented the same homeobox com- Fig. 8. This work was supported by grants from the prising a region of ~1.7 kb. It was named Iroquois-homeo- Deutsche Forschungsgemeinschaft (SFB No. 271) and box-3 (Irx3). To expand the Irx3 cDNA fragment a further from DGICYT to J. Modolell, by Fundacion Ramon Areces, 5′, an E14.5 randomly primed cDNA library (kindly pro- by the Leibnitz program and in part by the European Union. vided by Dr. R. Wehr) was screened at high stringency using a5′-probe excluding homeobox sequences. One clone extended the previously isolated cDNA around 700 bp in References the 5′ direction. The Irx3 cDNA has a total length of ~2.4 kb including an open reading frame encoding a 507 amino acid Allende, M.L., Weinberg, E.S., 1994. The expression pattern of two zebra- protein. The sequence directly upstream of the proposed fish achaete-scute homolog (ash) genes is altered in the embryonic brain initiation methionine agreed in all six bases with the of the cyclops mutant. Dev. Biol. 166, 509–530. ‘Kozak consensus sequence’ (Kozak, 1987). Alvarez-Bolado, G., Swanson, L.W., 1996. Developmental Brain Maps: Another set of four cDNA clones contained fragments of Structure of the Emryonic Rat Brain. Elsevier, Amsterdam. the Irx1 gene and spanned a length of ~1.5 kb. The remain- Auffray, C., Rougeon, F., 1980. Purification of mouse immunoglobulin heavy-chain messenger RNAs from total myeloma tumor RNA. Eur. ing three cDNAs belonged to the Irx2 gene comprising in J. Biochem. 107, 303–314. total ~2.2 kb. The translation start points of Irx1 and Irx2 Bellefroid et al., 1997. EMBO J., in press. have not yet been cloned. Campuzano, S., Modolell, J., 1992. Patterning of the Drosophila nervous system: the achaete-scute gene complex. Trends Genet. 8, 202–208. 4.2. Northern blot analysis Chitnis, A., Kintner, C., 1996. Sensitivity of proneural genes to lateral inhibition affects the pattern of primary neurons in Xenopus embryos. Development 122, 2295–2301. RNAs were extracted from E10.5–E17.5 NMRI embryos Dealy, C.N., Clarke, K., Scranton, V., 1996. Ability of FGFs to promote using the lithium chloride-urea method described by (Auf- the outgrowth and proliferation of limb mesoderm is dependent on IGF- fray and Rougeon, 1980). PolyA + RNA was isolated from I activity. Dev. Dyn. 206, 463–469. total RNA using columns containing Poly(U) sepharose4B Fekete, D.M., 1996. Cell fate specification in the inner ear. Curr. Opin. + Neurobiol. 6, 533–541. (Pharmacia). PolyA RNA (3 mg) was separated in a 1.2% Ferreiro, B., Kintner, C., Zimmerman, K., Anderson, D., Harris, W.A., agarose-formaldehyde gel and transferred to a nylon mem- 1994. XASH genes promote neurogenesis in Xenopus embryos. brane (Qiagen). Hybridizations with the Irx1, -2 and -3 Development 120, 3649–3655. specific probes were performed overnight at 42°C in 50% Ghysen, A., Dambly-Chaudiere, C., Jan, L.Y., Jan, Y.N., 1993. Cell inter- formamide, 1M sodium chloride, 1% SDS and 100 mg/ml actions and gene interactions in peripheral neurogenesis. Genes Dev. 7, × 723–733. salmon sperm DNA. Blots were sequentially washed in 2 Goddard, J.M., Rossel, M., Manley, N.R., Capecchi, M.R., 1996. Mice SSC, 0.5% SDS (two times 30 min at 65°C) and in 0.1× with targeted disruption of Hoxb-1 fail to form the motor nucleus of SSC, 0.5%SDS (once for 30 min at 65°C). the VIIth nerve. Development 122, 3217–3228. A. Bosse et al. / Mechanisms of Development 69 (1997) 169–181 181

Gomez-Skarmeta, J.L., Modolell, J., 1996. araucan and caupolican provide Noden, D.M., Van de Water, T.R., 1992. Genetic analyses of mammalian a link between compartment subdivisions and patterning of sensory ear development. Trends Neurosci. 15, 235–237. organs and veins in the Drosophila wing. Genes Dev. 10, 2935–2945. Ohuchi, H., Shibusawa, M., Nakagawa, T., Ohata, T., Yoshioka, H., Hirai, Go´mez-Skarmeta et al., 1997. EMBO J., in press. Y., Nohno, T., Noji, S., Kondo, N., 1997. A chick wingless mutation Guillemot, F., 1995. Analysis of the role of basic-helix-loop-helix tran- causes abnormality in maintenance of Fgf8 expression in the wing api- scription factors in the development of neural lineages in the mouse. cal ridge, resulting in loss of the dorsoventral boundary. Mech. Dev. 62, Biol. Cell 84, 3–6. 3–13. Guillemot, F., Joyner, A.L., 1993. Dynamic expression of the murine Oliver, G., Wehr, R., Jenkins, N.A., Copeland, N.G., Cheyette, B.N., Achaete-Scute homologue Mash-1 in the developing nervous system. Hartenstein, V., Zipursky, S.L., Gruss, P., 1995. Homeobox genes and Mech. Dev. 42, 171–185. connective tissue patterning. Development 121, 693–705. Hadrys et al., 1997. Development 125, in press. Parks, A.L., Huppert, S.S., Muskavitch, M.A., 1997. The dynamics of Hui, C.C., Slusarski, D., Platt, K.A., Holmgren, R., Joyner, A.L., 1994. neurogenic signalling underlying bristle development in Drosophila Expression of three mouse homologs of the Drosophila segment polarity melanogaster. Mech. Dev. 63, 61–74. gene cubitus interruptus, Gli, Gli-2, and Gli-3, in ectoderm- and meso- Rivolta, M.N., 1997. Transcription factors in the ear: molecular switches derm-derived tissues suggests multiple roles during postimplantation for development and differentiation. Audiol. Neuro-Otol. 2, 36–49. development. Dev. Biol. 162, 402–413. Rowitch, D.H., McMahon, A.P., 1995. Pax-2 expression in the murine Jan, Y.N., Jan, L.Y., 1994. Neuronal cell fate specification in Drosophila. neural plate precedes and encompasses the expression domains of Curr. Opin. Neurobiol. 4, 8–13. Wnt-1 and En-1. Mech. Dev. 52, 3–8. Joyner, A.L., 1996. Engrailed, Wnt and regulate midbrain-hind- Ruben, R.J., Rapin, I., 1980. Plasticity of the developing auditory system. brain development. Trends Genet. 12, 15–20. Ann. Otol. Rhinol. Laryngol. 89, 303–311. Kageyama, R., Sasai, Y., Akazawa, C., Ishibashi, M., Takebayashi, K., Ruben, R.J., Van De Water, T.R., Rubel, E., 1986. The Biology of Change Shimizu, C., Tomita, K., Nakanishi, S., 1995. Regulation of mammalian in Otolaryngology. Elsevier, Clearwater Beach, FL. neural development by helix-loop-helix transcription factors. Crit. Rev. Schimmang, T., Lemaistre, M., Vortkamp, A., Ruther, U., 1992. Expres- Neurobiol. 9, 177–188. sion of the zinc finger gene Gli3 is affected in the morphogenetic mouse Keynes, R., Krumlauf, R., 1994. Hox genes and regionalization of the mutant extratoes (Xt). Development 116, 799–804. nervous system. Annu. Rev. Neurosci. 17, 109–132. Sommer, L., Shah, N., Rao, M., Anderson, D.J., 1995. The cellular func- Kozak, M., 1987. An analysis of 5’-noncoding sequences from 699 verte- tion of MASH1 in autonomic neurogenesis. Neuron 15, 1245–1258. brate messenger RNAs. Nucleic Acids Res. 15, 8125–8148. Swanson, G.J., Howard, M., Lewis, J., 1990. Epithelial autonomy in the Krumlauf, R., 1994. Hox genes in vertebrate development. Cell 78, 191– development of the inner ear of a bird embryo. Dev. Biol. 137, 243–257. 201. Tanabe, Y., Jessell, T.M., 1996. Diversity and pattern in the developing Lee, J.E., 1997. NeuroD and neurogenesis. Dev. Neurosci. 19, 27–32. spinal cord. Science 274, 1115–1123. Leyns, L., Go´mez-Skarmeta, J.L., Dambly-Chaudiere, C., 1996. iroquois: a Torres, M., Gomez-Pardo, E., Gruss, P., 1996. Pax2 contributes to inner prepattern gene that controls the formation of bristles on the thorax of ear patterning and optic nerve trajectory. Development 122, 3381–3391. Drosophila. Mech. Dev. 59, 63–72. Treisman, J., Gonczy, P., Vashishtha, M., Harris, E., Desplan, C., 1989. A Lo, L., Guillemot, F., Joyner, A.L., Anderson, D.J., 1994. MASH-1: a single amino acid can determine the DNA binding specificity of home- marker and a mutation for mammalian neural crest development. Per- odomain proteins. Cell 59, 553–562. spect. Dev. Neurobiol. 2, 191–201. Vervoort, M., Dambly-Chaudiere, C., Ghysen, A., 1997. Cell fate deter- Lumsden, A., Krumlauf, R., 1996. Patterning the vertebrate neuraxis. mination in Drosophila. Curr. Opin. Neurobiol. 7, 21–28. Science 274, 1109–1115. Vortkamp, A., Gessler, M., Grzeschik, K.H., 1991. GLI3 zinc-fingel gene Mansour, S.L., Thomas, K.R., Capecchi, M.R., 1988. Disruption of the interrupted by translocations in Greig syndrome families. Nature 352, protooncogene int-2 in mouse embryo-derived stem cells: a general 539–540. strategy for targeting mutations to non-selectable genes. Nature 336, Wilkinson, D.G., Bhatt, S., McMahon, A.P., 1989. Expression pattern of 348–352. the FGF-related proto-oncogene int-2 suggests multiple roles in fetal Mansouri, A., Hallonet, M., Gruss, P., 1996. Pax genes and their roles in development. Development 105, 131–136. cell differentiation and development. Curr. Opin. Cell Biol. 8, 851–857. Wilkinson, D.G., Nieto, M.A., 1993. Detection of messenger RNA by in McGinnis, W., Krumlauf, R., 1992. Homeobox genes and axial patterning. situ hybridization to tissue sections and whole mounts. Methods Cell 68, 283–302. Enzymol. 225, 361–373. McNeill, H., Yang, C.H., Brodsky, M., Ungos, J., Simon, M.A., 1997. Wilkinson, D.G., Peters, G., Dickson, C., McMahon, A.P., 1988. Expres- mirror encodes a novel PBX-class homeoprotein that functions in the sion of the FGF-related proto-oncogene int-2 during gastrulation and definition of the dorsal-ventral border in the Drosophila eye. Genes Dev. neurulation in the mouse. EMBO J. 7, 691–695. 11, 1073–1082. Williams, P.L., Warwick, R., Dyson, M., Bannister, L.H., 1989. Gray’s Moscoso del Prado, J., Garcia-Bellido, A., 1984. Genetic regulation of the Anatomy. Churchill Livingstone. Acheate-scute complex of Drosophila melanogaster. Roux’s Arch. Dev. Zimmerman, K., Shih, J., Bars, J., Collazo, A., Anderson, D.J., 1993. Biol. 193, 242–245. XASH-3, a novel Xenopus achaete-scute homolog, provides an early Muskavitch, M.A., 1994. Delta-notch signaling and Drosophila cell fate marker of planar neural induction and position along the mediolateral choice. Dev. Biol. 166, 415–430. axis of the neuraj plate. Development 119, 221–232.